Development of a Rapid Global
Aircraft Emissions Estimation Tool
with Uncertainty Quantification
Nicholas Simone
An MIT SM thesis supervised by Prof. Steven Barrett
Date:
Report No:
Website:
January 2013
LAE-2013-002-T
LAE.MIT.EDU
Development of a Rapid Global Aircraft Emissions
Estimation Tool with Uncertainty Quantification
by
Nicholas W. Simone
B.S. Mechanical Engineering
Worcester Polytechnic Institute, 2008
SUBMITTED TO THE DEPARTMENT OF AERONAUTICS AND ASTRONAUTICS IN
PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF SCIENCE IN AERONAUTICS AND ASTRONAUTICS
AT THE
MASSACHUSETTS INSTITUTE OF TECHNOLOGY
February 2013
© 2013 Massachusetts Institute of Technology. All rights reserved.
Signature of Author………………………………………………………………....................................
Department of Aeronautics and Astronautics
January 29, 2013
Certified by………………………………………………………………………………………………………….....
Steven R.H. Barrett
Assistant Professor of Aeronautics and Astronautics
Thesis Supervisor
Accepted by………………………………………………...………………………………………………...........
Eytan H. Modiano
Professor of Aeronautics and Astronautics
Chair, Graduate Program Committee
[Page Intentionally Left Blank]
2
Development of a Rapid Global Aircraft Emissions Estimation
Tool with Uncertainty Quantification
by
Nicholas W. Simone
Submitted to the Department of Aeronautics and Astronautics
On January 29, 2013 in Partial Fulfillment of the
Requirements for the Degree of Master of Science in
Aeronautics and Astronautics
at the Massachusetts Institute of Technology
ABSTRACT
Aircraft emissions impact the environment by changing the radiative balance of the atmosphere
and impact human health by adversely affecting air quality. Many tools used to quantify
aircraft emissions are not open source and in most cases are computationally expensive. This
limits their usefulness for studies that require rapid simulation, such as uncertainty
quantification and assessment of many policy options. We describe the methods used to
develop the open source Aviation Emissions Inventory Code (AEIC) and produce a global
emissions inventory for the year 2005 from scheduled civil aviation, with quantified uncertainty.
This is the most up-to-date openly available inventory for use in atmospheric modeling studies.
We estimate that in 2005, scheduled civil aviation was responsible for 180.6 Tg (90% CI: 136.1232.9 Tg) of fuel burn, equating to 155.5 Tg of CO2 as C (90% CI: 117.3-200.7 Tg) and 0.108 Tg of
SOx as S (90% CI: 0.080-0.142 Tg) emissions. 2.689 Tg of NOx as NO2 (90% CI: 1.761-3.804 Tg),
0.749 Tg of CO (90% CI: 0.422-1.145 Tg), and 0.201 Tg of HC as CH4 (90% CI: 0.072-0.362 Tg)
were also emitted. 92% of fuel burn took place in the northern hemisphere. Landing and
takeoff operations were responsible for 9.1% of total global fuel burn, while 70.6% of fuel burn
occurred above 8 km. Our total fuel burn estimate agrees within 4% of other published
emissions inventories for the years 2004 and 2006, which is within the uncertainty range of the
analysis.
Thesis Supervisor: Steven Barrett
Title: Assistant Professor of Aeronautics and Astronautics
3
[Page Intentionally Left Blank]
4
Acknowledgements
I would first like to thank the General Electric Company for providing the funding for me
to complete my graduate work through the Advanced Courses in Engineering (ACE)
program, under the supervision of Ken Gould. I am extremely grateful for this financial
support.
I would like to thank my advisor, Professor Steven Barrett, for his guidance throughout
my research. From the very start, he had a vision of what this project could become and
the impact it could have, both of which were instrumental in how my research turned
out. He was always available to give comments or suggestions, despite his busy schedule
and other projects. I am also grateful for his understanding of my work schedule and
priorities while I completed my research; my experience would not have been the same
without this.
Thank you to Marc Stettler for his help and guidance throughout the development of
v2.0 of AEIC. His experience with the LTO cycle and aircraft emissions has been
invaluable throughout my research.
I must also thank Tom Reynolds, of Lincoln Labs, for his help and guidance on the
operational aspects of air travel. His knowledge and work has been invaluable
throughout this project.
I want to thank all of my GE colleagues for their support while I completed my degree.
They were very accommodating of my school schedule and the workload I had to
balance.
Thanks to everyone at the PARTNER/LAE/ICAT lab for the knowledge and insight that
both enhanced my research and personal education while at MIT. In particular, I would
like to thank Seb Eastham, Akshay Ashok, and Steve Yim for their help with servers, code,
and helping me understand the downstream use of the results of my research.
I would like to thank my parents, Tara and Keith, and my sister, Amanda, for their
continued love and support.
Last, but definitely not least, I want to thank my girlfriend, Heidi, for her love,
encouragement, and understanding while I completed my degree. I know the late nights
and never-ending to do lists were tough, but I would not have made it through without
you! Thank you for everything. I love you so very much.
5
[Page Intentionally Left Blank]
6
Table of Contents
1.
Introduction ........................................................................................................... 15
1.1. Context ............................................................................................................. 15
1.2. Purpose ............................................................................................................ 16
2.
Methods ................................................................................................................. 17
2.1. Landing and takeoff (LTO) cycle ........................................................................ 17
2.2. Flight scheduling ............................................................................................... 18
2.3. Aircraft fuel burn .............................................................................................. 18
2.3.1.
Flight tracks ................................................................................................... 18
2.3.2.
Aircraft performance ..................................................................................... 19
2.4. Emissions .......................................................................................................... 20
2.5. Correction for operational inefficiencies .......................................................... 21
2.6. Uncertainty quantification ................................................................................ 22
3.
Year 2005 Emissions Inventory Results ................................................................... 24
3.1. Code performance ............................................................................................ 24
3.2. Worldwide totals .............................................................................................. 24
3.3. Spatial distribution ........................................................................................... 25
3.3.1.
Global ............................................................................................................ 25
3.3.2.
Latitude ......................................................................................................... 26
3.3.3.
Longitude ...................................................................................................... 27
3.3.4.
Altitude ......................................................................................................... 28
3.4. Fuel breakdown ................................................................................................ 29
3.4.1.
By country of origin/destination .................................................................... 29
3.4.2.
By country land border .................................................................................. 32
3.5. Comparison to other inventories ...................................................................... 33
4.
Case Studies ........................................................................................................... 35
4.1. Combustor water injection as NO x abatement for takeoff operations .............. 35
4.1.1.
Background ................................................................................................... 35
4.1.2.
Methodology ................................................................................................. 36
4.1.3.
Results .......................................................................................................... 37
7
4.2. Air traffic management system efficiency ......................................................... 38
4.2.1.
Background ................................................................................................... 38
4.2.2.
Methodology................................................................................................. 39
4.2.3.
Results .......................................................................................................... 39
5.
Conclusions ............................................................................................................ 41
6.
References ............................................................................................................. 44
8
List of Figures
Figure 1: Column sum of global fuel burn from scheduled civil aviation for the year 2005.
...................................................................................................................................... 25
Figure 2: Latitudinal distribution of global emissions inventories using AEIC for the year
2005 (blue, this thesis) and published AEDT results by Wilkerson et al. (2010) (red). .... 26
Figure 3: Longitudinal distribution of global emissions inventories using AEIC for the year
2005 (blue, this thesis) and published AEDT results by Wilkerson et al. (2010) (red). .... 27
Figure 4: Altitudinal distribution of global emissions inventories using AEIC for the year
2005 (blue, this thesis) and published AEDT results by Wilkerson et al. (2010) (red). .... 28
9
List of Tables
Table 1: Summary of global emissions ........................................................................... 25
Table 2: Total fuel burn by country of origin/destination in 2005, averaged. ................. 29
Table 3: Per capita fuel burn and CO2 emissions by country of origin/destination in 2005,
averaged. Countries with populations of under 1 million have been omitted. ............... 30
Table 4: Fuel burn breakdown for EU ............................................................................ 31
Table 5: Fuel burn within country land borders. Left: The ten countries with the largest
amount of absolute fuel burn in 2005 (Tg). Right: The ten countries with the highest fuel
burn density in 2005 (kg/km2). ...................................................................................... 32
Table 6. Comparison of published global emissions inventories for scheduled civil
aviation.......................................................................................................................... 33
Table 7. Results of water injection study. ...................................................................... 37
Table 8. ATM system improvement opportunity for different emissions ....................... 40
Table 9. Contribution of each flight phase to inefficiency for different emissions ......... 40
10
[Page Intentionally Left Blank]
11
List of Acronyms
AEDT
Aviation Environmental Design Tool
AEIC
Aviation Emissions Inventory Code
AFL
Above Field Level
ATC
Air Traffic Control
ATM
Air Traffic Management
BADA
Base of Aircraft Data
BC
Black Carbon
BFFM
Boeing Fuel Flow Method 2
CANSO
Civil Air Navigation Services Organization
CH4
Methane (used for mass basis)
CO
Carbon Monoxide
CO2
Carbon Dioxide
EI
Emission Index
EU
European Union
GEOS-5
Goddard Earth Observing System Model, Version 5
H 2O
Water (Typically as Water Vapor)
HC
Unburned Hydrocarbons
ICAO
International Civil Aviation Organization
kg
Kilogram (103 grams)
km
Kilometer (103 meters)
LTO
Landing and Takeoff
NASA
National Aeronautics and Space Administration
NM
Nautical Mile
NO2
Nitrogen Dioxide (used for mass basis)
12
NOx
Oxides of Nitrogen
OAG
Official Airline Guide
OC
Organic Carbon
QUANTIFY
Quantifying the Climate Impact of Global and European Transport Systems
RVSM
Reduced Vertical Separation Minimum
SFC
Specific Fuel Consumption
SOx
Oxides of Sulfur
SVI
SO3 and SO4
TAS
True Air Speed
Tg
Teragram (1012 grams)
T/O
Takeoff
TIM
Time In Mode
UK
United Kingdom
13
[Page Intentionally Left Blank]
14
1. Introduction
1.1.
Context
Aviation is currently responsible for approximately 3% of global fossil fuel consumption
(IEA/OECD, 2007) and 12% of transportation related CO2 emissions (ICAO, 2010). Global
aviation traffic has grown substantially over the last several decades and is expected to
continue increasing: passenger travel has increased ten-fold since 1970, doubled since
1995 (Airbus, 2012), and long term forecasts from ICAO place growth rates at up to 6.2%
per year (ICAO, 2012). Emissions from aircraft consist of CO2, CO, NOx, H2O, SOx,
unburned hydrocarbons (HC), black carbon (BC), and organic carbon (OC) (Lee et al.,
2009). These emissions impact both air quality (causing adverse human health impacts)
and the climate at regional and global scales.
Emissions from aircraft differ from other anthropogenic emission sources in that the
vast majority occurs at high altitude (Olsen et al., 2012), with the exception of species
associated with low thrust operation (CO and HC). The altitude of the emissions can
cause a disproportionate increase in their effect on the climate, as in the case of NO x
(Gauss et al., 2006). Overall, aviation emissions make up approximately 3.5%-4.9% of
the total radiative forcing due to all anthropogenic emissions (Lee et al., 2009), although
significant uncertainties remain. In addition, more recent work has found that high
altitude aircraft emissions perturb surface air quality. Barrett et al. (2010a) estimated
that 80% of the ~10,000 premature mortalities per year due to the adverse air quality
impacts from aircraft emissions come from emissions at cruise altitudes. This represents
~1% of the estimated 800,000 premature deaths due to air pollution from
anthropogenic sources (Krzyzanowski & Cohen, 2008).
Due to the processes and chemical reactions that take place involving aircraft emissions,
we must rely on global atmospheric models to assess their effects, requiring 4-D (3-D
15
and time) quantification of the emissions. The models used to develop these emissions
datasets are typically high fidelity aircraft performance and emissions models, which
make them computationally expensive. Many times, they are not open source, reducing
the impact they can have in the research domain.
The computational intensity of models becomes important when rapid simulations are
needed to assess several scenarios, or to quantify uncertainty. Due to the complexity of
the systems being modeled and the lack of knowledge of the physical processes that
take place, there is significant uncertainty in both the emissions estimates (Stettler et al.,
2011; Lee et al., 2007) and the downstream effects of the emissions (Lee et al., 2009).
This high level of uncertainty makes it paramount that uncertainty quantification is
included as part of aviation impact studies. This is often computationally prohibitive
with current tools and previous estimates of the uncertainty in civil aviation emissions in
total have not previously been made.
1.2.
Purpose
We describe the methods used and the results obtained from the Aviation Emissions
Inventory Code (AEIC). AEIC was originally developed by Stettler et al. (2011) to quantify
emissions and associated uncertainty from landing and takeoff operations. We extend
the modeling domain to include the entire aircraft flight in order to quantify the global
emissions of scheduled civil aviation for the year 2005. We reduce the modeling
complexity through the utilization of assumptions in order to keep the computational
intensity low enough to allow for rapid simulation of annual global emissions, allowing
for uncertainty quantification through a Monte Carlo simulation. We produce the only
publically available aviation emissions inventory with an emissions year within the past
decade, and the first “bottom-up” estimate for the uncertainty in civil aviation emissions
as a whole.
16
2. Methods
In this section, we provide an overview of the methods and assumptions used to
calculate the global aircraft emissions inventory. When appropriate, computationally
efficient assumptions have been used to reduce the computational intensity, greatly
increasing the practicality of the model while maintaining its ability to adequately
capture the dynamics required for a global estimate. There are two distinct areas of
aircraft operations that are modeled: landing and takeoff (LTO) operations and non-LTO
operations (climb to cruise, cruise, and descent). LTO emissions are modeled per Stettler
et al. (2011) and are defined as those that take place between 0 and 3000 feet above
field level (AFL), consistent with their approach. Cruise operations are the main focus of
the methodology development and include operations above 3000 feet AFL. The areas
described herein consist of an LTO cycle overview, flight scheduling, aircraft fuel burn,
emissions calculations, corrections for operational inefficiencies in the system, and
uncertainty quantification.
2.1.
Landing and takeoff (LTO) cycle
We calculate LTO fuel burn and emissions using the methodology described in Stettler
et al. (2011). A brief overview is given here. The LTO cycle is defined using specific timesin-mode (TIMs) for different portions of the cycle. ICAO has defined a default LTO cycle
consisting of takeoff, climb, approach, and taxi/ground idle with specified thrust levels
and times for each portion (ICAO, 2008). As defined in Stettler et al. (2011), the ICAO
default cycle is typically not representative of real world operations. Thus, a more
representative cycle that consists of TIMs and thrust levels for taxi out, taxiway
acceleration, hold, takeoff, initial climb, climb out, approach, landing roll, reverse thrust,
and taxi in has been used.
17
2.2.
Flight scheduling
We use the Official Airline Guide (OAG) (OAG Aviation, 2005) to generate a schedule of
flights for the year 2005. The OAG contains only scheduled civil air traffic; no adjustment
is made to the resulting output for unscheduled or canceled flights. There is more
discussion on the effect of this assumption in section 3.5.
We model traffic from 2,572 airports around the world in order to capture 99% of the
passenger enplanements contained within the OAG. The OAG data is used to generate
unique aircraft-airport directional pairs and calculate the number of times each pair is
flown over a specified time period.
2.3.
Aircraft fuel burn
2.3.1. Flight tracks
Flight tracks for each unique aircraft-airport directional pair from the OAG data are
generated in order to calculate and track fuel burn and emissions for each flight. We
assume all aircraft follow a great circle path between the departure and arrival airports.
We reduce the absolute error introduced by using this assumption afterwards by
incorporating lateral inefficiency metrics available in the literature (discussed in section
2.5).
We incorporate wind data from GEOS-5 (Rienecker et al., 2008) into the analysis. This
data consists of wind direction and annually averaged wind speed. The incorporation of
this data serves to change the relationship between the true air speed (TAS) and ground
speed for an aircraft depending on spatial location and heading. Thus, flights with a
headwind component fly slower with respect to the ground and those with a tailwind fly
faster.
18
2.3.2. Aircraft performance
We calculate aircraft performance using EUROCONTROL’s BADA (Base of Aircraft Data)
Version 3.9 (Eurocontrol Experimental Center, 2011). BADA contains support for 338
total aircraft; 117 of which are “directly supported”, in that their performance and
operational characteristics are specifically modeled in BADA. The remaining 221 aircraft
are supported by similarity to the other models, as determined by EUROCONTROL.
Following the approach of Stettler et al. (2011), we estimate performance for aircraft
not specified by EUROCONTROL in BADA by modeling them as other similar aircraft.
To optimize calculations in this area, we calculate aircraft performance using a predefined look-up table, as opposed to a physics-based aircraft performance model, which
would be computationally expensive. There is a unique look-up table for each BADA
supported aircraft that includes TAS, fuel flow rate, and rate of climb/descent for
various flight levels and aircraft weights. This method allows fuel burn and velocity to be
calculated for each flight chord using only a table look-up.
To make calculations more efficient, we simulate each unique aircraft-airport directional
pair only once. Total fuel burn for a given interval is then calculated by multiplying the
output from the one simulated flight by the number of times that flight operates over
an interval. This allows us to estimate emissions for over 27 million flights annually using
~110,000 simulations (a 99.6% reduction), while still capturing average characteristics,
as will be shown in the results.
Assumptions for aircraft takeoff weight and cruise altitude are required because FDR
(flight data recorder) and radar track information are not used in the model. We utilize a
takeoff weight assumption from Eyers et al. (2004), which consists of the empty weight
of the airframe, 60.9% of maximum payload capacity, fuel payload to fly to the
destination, 5% extra reserve fuel, fuel for a diversion [100 NM (nautical miles) for short
19
haul and 200 NM for long haul], and fuel for a low altitude hold (45 min for short haul
and 30 min for long haul) (Eyers et al., 2004). We define short haul flights as flights less
than or equal to three hours in length and long haul flights as those greater than three
hours. Aircraft cruise altitude is nominally set to 7,000 feet (ISA pressure altitude) below
the maximum cruise altitude of the aircraft, as specified by BADA. The effects of both
the takeoff weight and cruise assumptions are accounted for using uncertainty
distributions (discussed later).
2.4.
Emissions
We calculate emissions for all flights based on the aircraft performance calculations and
the specific species emitted. For SOx emissions, we assume a mass fuel sulfur content
(FSC) of 600 ppm (Hileman et al., 2010; Stettler et al., 2011) and a 2% conversion
efficiency to SVI (Barrett et al., 2010b). For CO2, we utilize a constant emission index (EI)
of 3160 g-CO2/kg-fuel (Stettler et al., 2011). In this manner, both SOx and CO2 emissions
scale directly with fuel burn.
For NOx, HC, and CO, we utilize EIs from the ICAO Engine Emissions Databank (CAA,
2009), along with Boeing’s Fuel Flow Method 2 (BFFM2) (Baughcum et al., 1996) to
calculate the emissions for all flights. The ICAO databank contains information on
emissions and fuel flow from engines certified for flight at four different certification
thrust levels: 7%, 30%, 85%, and 100%. The ICAO data is supplied at sea level, engine
uninstalled conditions and adjustments are made for engine installation effects that
increase fuel flow at a given thrust and altitude effects, as suggested by Baughcum et al.
(1996).
BFFM2 provides a method to interpolate/extrapolate between the thrust points in the
databank, as well as extrapolate the data from sea level to altitude. The interpolation
20
between certification points consists of a log-log linear fit for NOx and a log-log bilinear
fit for HC and CO to certification measurements. Extrapolating the emissions indices to
altitude requires a correction on both the fuel flow rate from BADA, as well as a
correction to the reference EIs from the ICAO databank. The fuel flow correction
depends on ambient pressure, temperature, and flight Mach number. The result is a sea
level, static, standard day equivalent fuel factor that can be used with the ICAO
databank EIs. HC and CO EIs are then corrected for ambient temperature and pressure,
while the NOx EI is corrected for ambient temperature, pressure, and humidity level.
Consistent with other studies, we assume a relative humidity of 60% for the entire flight
(Baughcum et al., 1999; Kim et al., 2007; Pham et al., 2010).
2.5.
Correction for operational inefficiencies
We correct the non-LTO fuel burn and emissions calculations according to lateral
inefficiency factors (Reynolds, 2008, 2009). Reynolds (2008, 2009) examined several sets
of flight data to determine the average increase in ground track from great circle for
various flights due to factors such as route structure, air traffic control (ATC) procedures
and deviations for weather and congestion. We utilize only the portion referred to as
“Ground Track Extension”; that is, the added distance flown by an aircraft when
compared to leaving the terminal area in a straight line, flying enroute to the arrival
airport on a great circle path, and approaching the terminal area in a straight line. No
adjustments are made for inefficiencies due to less than optimum cruise altitude or
speed.
The lateral inefficiency factors serve to increase the amount of fuel burned and
emissions for a given flight. We incorporate lateral inefficiencies from the United States
and Europe. For areas other than the United States and Europe, lateral inefficiencies
21
from the United States are assumed. Departure and arrival inefficiencies are based on a
50 NM terminal area radius and enroute inefficiencies are with respect to a great circle,
making them applicable to this analysis.
From Reynolds (2008), average departure inefficiency is approximately 8-9 NM, while
average arrival inefficiencies are 27-28 NM due to vectoring and holding. Enroute
extension is 5-6% of great circle distance. It should be noted that there is significant
variability around these averages. As such, we only use these average inefficiencies for a
nominal simulation and model the distribution around the average in our uncertainty
assessment, which is discussed in the following section.
2.6.
Uncertainty quantification
We make use of an uncertainty approach similar to Stettler et al. (2011), with necessary
additions for the inclusion of cruise calculations into the analysis. We approximate
uncertainty distributions using a triangular distribution [specified herein by (min, mode,
max)] and quantify the level of uncertainty in each output by using a Monte Carlo
simulation consisting of 1000 model executions. We utilize magnitudes for LTO
operational (thrust levels, times in modes, etc.) and scientific (emissions indices)
uncertainties from Stettler et al. (2011).
The uncertainties we take into account for the cruise modeling are cruise altitude, takeoff weight, departure ground track extension, en-route ground track extension, arrival
ground track extension, aircraft drag, and aircraft engine specific fuel consumption.
Uncertainty ranges are discussed next.
Ground track extension uncertainties are based on data from Reynolds (2008); arrival
and departure uncertainties are distributions of the extra distance flown, while en-route
uncertainty is modeled with a multiplier on the nominal inefficiency. The departure
22
distributions are (0, 3, 20) NM for the US and (0, 5, 25) NM for the EU, while the arrival
distributions are (0, 2, 75) NM for the US and (0, 22, 57) NM for the EU. Enroute
multipliers are (0.25, 1, 2) for the US and (0.25, 1, 2.5) for the EU.
Variation in cruise altitude has been shown to be approximately 3000 feet for a 1σ
uncertainty level (Lee, 2005; Lee et al., 2007); we use (-6750, 0, 6750) feet to represent
the variation around our nominal cruise altitude assumption. We use a takeoff weight
multiplier distribution of (0.7075, 1, 1.2925) to model the uncertainty in takeoff weight,
which is representative of a 13% 1σ uncertainty (Lee et al., 2007).
The BADA performance model makes use of several simplifying assumptions that affect
the fuel burn calculations in our analysis. Two significant assumptions are related to
aircraft drag and engine specific fuel consumption (SFC). The modeling of SFC in BADA
does not fully capture the dependency of aircraft engine performance on altitude and
speed; thus, the 1σ uncertainty is approximately 11% (Lee, 2005; Yoder, 2007). We use a
multiplier on flight fuel burn with a range of (0.7525, 1, 1.2475) to capture the
uncertainty here. Similarly, for aircraft lift/drag performance, the BADA model does not
fully capture the dependency on altitude and speed, yielding a 1σ uncertainty level of
14% (Lee, 2005). Here we also use a multiplier on flight fuel burn; the range is (0.685, 1,
1.315). The uncertainty levels used for cruise altitude, takeoff weight, SFC, and drag are
consistent with those used in similar studies (Lee et al., 2007), although this has never
been attempted fleet-wide.
We note that the uncertainty magnitudes we have accounted for capture the variation
present in individual flights and will overestimate the uncertainty surrounding the fleetwide average, as the variation on an individual basis will bound the uncertainty of the
average. Results are quoted with 90% confidence intervals, unless otherwise specified,
based on 1000 executions of AEIC. Nominal results are obtained from a simulation using
nominal inputs.
23
3. Year 2005 Emissions Inventory Results
3.1.
Code performance
Fast model execution times allow global fleet-wide simulations to be utilized in ways
that have not been possible in the past, such as rapid policy analyses and fleet-wide
uncertainty quantification. AEIC is capable of generating global emissions for a full year
in approximately one hour on a single core and can be parallelized for multiple model
simulations.
3.2.
Worldwide totals
Table 1 contains a summary of worldwide emissions results, including uncertainties. We
estimate that global fuel burn from scheduled civil aviation is approximately 180.6 Tg
(90% CI: 136.1-232.9 Tg). The emissions with the largest uncertainty are HC emissions,
attributable to the large amount of uncertainty that exists in the EI for HC emissions at
low thrust. It should be noted that HC emissions also show the greatest amount of
variability among different emissions inventory studies (Olsen et al., 2012; Kim et al.,
2007). A simulation with no wind results in a fuel burn decrease of approximately 0.6%.
24
Table 1: Summary of global emissions
Nominal (Tg)
Mean (Tg)
Median (Tg)
Coefficient of
Variation
90% Confidence
Interval (Tg)
Fuel Burn
180.6
180.9
178.5
16.7%
136.1-232.9
CO2 as C
155.6
155.8
153.7
16.8%
117.3-200.7
SOx as S
0.108
0.108
0.107
17.9%
0.080-0.142
NOx as NO2
2.689
2.631
2.535
23.7%
1.761-3.804
CO
0.749
0.760
0.749
28.8%
0.422-1.145
HC as CH4
0.201
0.203
0.196
42.6%
0.072-0.362
Emission
3.3.
Spatial distribution
3.3.1. Global
Figure 1: Column sum of global fuel burn from scheduled civil aviation for the year
2005.
Figure 1 shows the spatial distribution of global fuel burn from scheduled civil aviation
for 2005. 44.5% of the globe has an annual fuel burn total of less than 1 kg/km2. We
note that the lack of track dispersion when generating the fuel burn totals will result in
25
more concentrated emissions than if the tracks are more spread out. In our analysis, we
have assumed that each aircraft flies the same great circle route between airports, while
actual flight tracks will have a distribution around this path due to separation
requirements, weather, etc.
3.3.2. Latitude
Figure 2: Latitudinal distribution of global emissions inventories using AEIC for the
year 2005 (blue, this thesis) and published AEDT results by Wilkerson et al. (2010)
(red).
Figure 2 contains the latitudinal distribution of fuel burn. 92% of fuel burn takes place in
the northern hemisphere, with 67% percent of global fuel burn taking place in the
northern mid-latitudes between 30°N and 60°N. Emissions all but cease lower than 45°S,
with 0.06% of fuel burn occurring there. The largest peak occurs between 40-41°N. This
peak is the result of three of the US’s busiest airports being in this area (John F. Kennedy
26
International Airport, Newark Liberty International Airport, and LaGuardia International
Airport). A comparison with AEDT results is also given.
3.3.3. Longitude
Figure 3: Longitudinal distribution of global emissions inventories using AEIC for
the year 2005 (blue, this thesis) and published AEDT results by Wilkerson et al.
(2010) (red).
Figure 3 contains a plot of the longitudinal distribution of fuel burn. There are three
distinct peaks in the plot, corresponding to the three heaviest traffic areas in Figure 1.
The peak between 120°W and 60°W accounts for 32.1% of global fuel burn and is largely
a result of North American air traffic. The peak from 15°W to 30°E is a result of
European air traffic, and accounts for 19.3% of global fuel burn. The last peak from 90°E
to 150°E contains 21.2% of global fuel burn and is a result of the traffic in East Asia and
Australia. A comparison with AEDT results is also given.
27
3.3.4. Altitude
Figure 4: Altitudinal distribution of global emissions inventories using AEIC for the
year 2005 (blue, this thesis) and published AEDT results by Wilkerson et al. (2010)
(red).
Figure 4 shows the distribution of fuel burn with altitude. Emissions from LTO
movements comprise approximately 9.1% of total global fuel burn. 70.6% of global fuel
burn occurs at altitudes greater than 8 km. A comparison with AEDT results is also given.
28
3.4.
Fuel breakdown
3.4.1. By country of origin/destination
Table 2 lists the ten countries with the highest fuel burn, based on the
origin/destination of flights. The fuel burn totals for flights originating in and arriving in
each country have been averaged. The proportion of fuel burn for domestic flights and
international flights is also included. LTO fuel burn for all flights is counted as domestic.
The United States has the largest fuel burn at 59.1 Tg (32.7% of the global total). It also
has the highest percentage of domestic traffic (71%). Hong Kong has the largest
percentage of fuel burn from international flights, with 94%.
Table 2: Total fuel burn by country of origin/destination in 2005, averaged.
Fuel Burn
(Tg)
59.1
% Of Global
Total
32.7%
% Domestic
% International
71%
29%
Japan
9.7
5.4%
40%
60%
United Kingdom
9.4
5.2%
13%
87%
China (excluding Hong Kong)
8.5
4.7%
63%
37%
Germany
6.7
3.7%
15%
85%
France
5.4
3.0%
17%
83%
Australia
4.4
2.4%
42%
58%
Canada
4.1
2.3%
45%
55%
Spain
3.9
2.2%
35%
65%
Hong Kong
3.5
2.0%
6%
94%
Country
United States of America
29
Table 3 contains data for the ten countries with the greatest per capita fuel burn and
CO2 emissions. These numbers are based on 2005 population statistics for each country
from the United Nations (United Nations, 2011). Countries with populations under 1
million have been omitted. The United Arab Emirates and Singapore have the highest
per capita fuel burn, both equating to approximately 2.4 tonnes of CO 2 per person. We
note that, particularly for city-states, this is not appropriately interpreted as aviation
CO2 emissions attributed to the average resident of each country due to international
passengers and visitors.
Table 3: Per capita fuel burn and CO2 emissions by country of origin/destination in
2005, averaged. Countries with populations of under 1 million have been omitted .
Fuel Burn/Person
(kg/person)
764
Tonne-CO2/Person
Singapore
755
2.4
Hong Kong
518
1.6
New Zealand
255
0.8
Australia
216
0.7
United States of America
199
0.6
Netherlands
178
0.6
Cyprus
166
0.5
Mauritius
159
0.5
United Kingdom
156
0.5
Country
United Arab Emirates
30
2.4
Table 4 shows the fuel burn breakdown for the 27 member states of the current
European Union (EU) (European Union, 2012). The largest contributor to the total EU
fuel burn is the international-non LTO phase of flights, which accounts for about 73% of
EU-attributed fuel burn. The total fuel burn for the EU is 59.8 Tg, which accounts for
33.1% of the global total – approximately equal to the United States. On a per capita
basis, the EU as a whole consumes 121 kg/person of fuel (0.4 tonne-CO2/person), which
would place it below the Top 10 countries in terms of per capita fuel burn (Table 3).
Table 4: Fuel burn breakdown for EU
Fuel Burn
(Tg)
13.0
Contribution To
Total
22%
International-Non LTO
43.7
73%
LTO
3.1
5%
Domestic-Non LTO
31
3.4.2. By country land border
Table 5 lists the ten countries with the highest level of absolute fuel burn and the
highest fuel burn density (fuel burn per unit area) within their land borders. We have
utilized the Gridded Population of the World for country land territory (CIESIN/Columbia
Univerity/CIAT, 2005). Six of the ten countries with the largest absolute fuel burn are
the six largest countries in the world (United States of America, China, Russia, Canada,
Australia, and Brazil). The countries with the highest fuel burn density are mostly
European countries that are relatively small in size. Germany, France, Japan, and the
United Kingdom appear in the top ten list of both absolute fuel burn and fuel burn
density.
Table 5: Fuel burn within country land borders. Left: The ten countries with the
largest amount of absolute fuel burn in 2005 (Tg). Right: The ten countries with
the highest fuel burn density in 2005 (kg/km2).
Fuel Burn
(Tg)
41.1
% Of Global
Total
22.8%
Belgium
Fuel Burn Density
(kg/km2)
14,755
China (excluding Hong Kong)
9.4
5.2%
Germany
10,713
Russia
6.9
3.8%
Switzerland
10,707
Canada
6.0
3.3%
United Kingdom
10,683
Germany
3.8
2.1%
Netherlands
10,427
France
3.7
2.1%
Japan
8,551
Japan
3.5
1.9%
United Arab Emirates
8,024
United Kingdom
2.9
1.6%
Korea
7,774
Australia
2.9
1.6%
Austria
7,112
Brazil
2.7
1.5%
France
6,695
Country
United States of America
32
Country
3.5.
Comparison to other inventories
Here we provide a brief comparison to other published global inventories for years close
to 2005: NASA/Boeing for 1999 (Sutkus Jr. et al., 2001; Olsen et al., 2012), Quantifying
the Climate Impact of Global and European Transport Systems (QUANTIFY) for 2000
(Owen et al., 2010; Olsen et al., 2012), AERO2k for 2002 (Eyers et al., 2004), and the
Aviation Environmental Design Tool (AEDT) for 2004 and 2006 (Wilkerson et al., 2010).
Table 6 contains a comparison of global fuel burn, NOx, CO, and HC for all of the
inventories. CO2 and SOx have intentionally been omitted, as they are directly
proportional to fuel burn (the latter depending on the FSC assumption).
Table 6. Comparison of published global emissions inventories for scheduled civil
aviation.
NASA/Boeing
Year 1999
QUANTIFY
Year 2000
AERO2k
Year 2002
AEDT
Year 2004
Fuel Burn (Tg)
136
152
156
174.0
AEIC
Year 2005
(This Thesis)
180.6
NOx as NO2 (Tg)
1.38
1.98
2.06
2.456
2.689
2.656
CO (Tg)
0.667
----
0.507
0.628
0.749
0.679
HC as CH4 (Tg)
0.226
----
0.063
0.090
0.201
0.098
Emission
AEDT
Year 2006
188.2
Our total fuel burns agrees within 4% of the inventories generated by AEDT for the years
2004 and 2006, which is well within the 90% confidence interval we calculated. AEDT is
a higher fidelity tool that incorporates radar track data and models each flight
individually, allowing it to account for actual flight paths and unscheduled/cancelled
flights. The NASA/Boeing, QUANTIFY, and AERO2k inventories agree with a general
trend of increasing fuel burn each year. No inventories for the year 2005 are available
for a direct comparison; however, Wilkerson et al. (2010) does qualitatively show that
fuel burn for 2005 was greater than both 2004 and 2006 (Wilkerson et al., 2010).
Unscheduled flights have been estimated to account for approximately 9% of global
33
flights annually (Kim et al., 2007), although their impact on fuel burn/emissions has not
been directly quantified.
The largest relative difference between our results and the other inventories are in the
CO and HC emissions. This is not unexpected, given the relatively large uncertainty in
the emission of these species and their sensitivity to power setting. Based on empirical
observations, we have assumed a lower thrust level for the LTO cycle than is typically
used, which increases both CO and HC emissions (Stettler et al., 2011). However, our
results are within 11.1% of the NASA/Boeing inventory and the relatively low CO and HC
emissions from AEDT have been noted during its development (Kim et al., 2007).
For comparison of spatial distribution, we have shown our results alongside the results
from AEDT 2006 in Figure 2-Figure 4 (latitudinal, longitudinal, and altitudinal) on the
basis that the AEDT inventory contains the most detail related to flight altitude and
location by using radar track data. In general, our results capture all of the same peaks
and valleys, but are of a lower magnitude (as expected after comparing the global
totals). The altitude distribution is slightly different due to the fact that our analysis has
not directly incorporated radar tracks or flight data.
34
4. Case Studies
4.1.
Combustor water injection as NOx abatement for takeoff operations
4.1.1. Background
Here, we perform a technology assessment using AEIC. We look at the total potential
benefit of using combustor water injection to lower takeoff NO x emissions. A combustor
water injection system has been chosen over a compressor misting system due to
feasibility (Daggett et al., 2007). NOx production in aircraft engines is closely linked to
combustor flame temperature (Daggett et al., 2010) and increases exponentially as
thrust setting increases (Baughcum et al., 1996). When water is injected into the
combustor, it lowers the flame temperature and results in lower NO x emissions;
estimated magnitudes are about an 80% reduction in NOx for a 1:1 water:fuel injection
ratio (Daggett et al., 2007).
Two potential system issues with water injection systems are corrosion (if purified water
is not used) and water freezing (Daggett et al., 2007). Operationally, if payload is not
decreased, the weight of the water and the storage/distribution system requires more
fuel per flight. In addition, the lower combustor temperature lowers the thermal
efficiency of the engine. This could result in an SFC increase of 2.0% when the water
injection system is being used (Daggett et al., 2010). It is the trade-off between the
added weight/lower engine efficiency and a reduction NOx EI that we wish to study.
35
4.1.2. Methodology
We utilize the results of a study completed by Daggett et al. (2007), where a combustor
water injection system for a 747-400ER was investigated. The water injection system
designed weighed approximately 750 lb. and the aircraft required 3340 lb. of water for a
1:1 water:fuel injection ratio during takeoff. To extend these results to all aircraft and
determine the impact on global emissions, we make the following assumptions:
1. Water injection is used on all flights during the takeoff, initial climb, and climb
out phases.
2. The water injection system weight scales linearly with aircraft empty weight.
3. For the LTO phase, the added weight due to the system and water can be
represented with an increase in thrust. For the non-LTO phase, we model the
added weight of the water injection system by increasing the empty weight of
the aircraft.
4. The SFC increase of 2.0% is applicable to all aircraft when the water injection
system is being utilized.
Based on the empty weight of the 747-400ER (from BADA), the water injection system
represents an increase in empty weight of approximately 0.2%; this fraction is used for
all aircraft.
To determine the thrust increase required for the weight of the water and system on
the 747-400ER, we first determined the average takeoff weight for all 747-400ER flights,
utilizing the takeoff weight assumption described in Section 2.3.2. Then, assuming a
linear thrust increase between this nominal takeoff weight and the 747-400ER
maximum takeoff weight (90% thrust for nominal weight per Stettler et al. (2011)
assumption and 100% thrust for maximum weight) (British Airways/IATA, 2002), we
calculate the required thrust increase to be approximately 0.2%. We utilize this thrust
increase for all aircraft during the flight phases mentioned in assumption 1.
36
4.1.3. Results
Table 7 contains the results of the water injection simulations. LTO and global results
with and without the SFC penalty described in Daggett et al. (2010) are tabulated. The
global potential for NOx reduction using this technology is 4.4%, with a potential for an
extra 0.1% reduction if the SFC penalty is reduced. The benefit for LTO NOx reduction is
much larger at 59.4-59.7%. This represents a substantial reduction in the NOx emitted in
the direct vicinity of the airport.
The global fuel burn penalty for this technology is 0.1-0.2%, but the LTO fuel burn
penalty could be up to 1.0%. The majority of the LTO fuel burn penalty is from the SFC
reduction. The fuel penalty due to the added weight is 0.1%, which is only 10% of the
total LTO fuel burn penalty. Because LTO fuel burn accounts for about 9% of total fuel
(Section 3.3.4), the takeoff SFC increase has a smaller relative effect on global fuel burn
than it does on LTO fuel burn.
Table 7. Results of water injection study.
LTO Fuel Burn
Change
LTO NOx
Change
Global Fuel Burn
Change
Global NOx
Change
Weight Penalty Only
(No SFC Penalty)
+0.1%
-59.7%
+0.1%
-4.5%
Weight + T/O SFC Penalty
+1.0%
-59.4%
+0.2%
-4.4%
Scenario
37
4.2.
Air traffic management system efficiency
4.2.1. Background
The diffusion of new aircraft technology is typically associated with long time constants.
The average lifetime of one aircraft is on the order of 25-30 years (ICAO,2007) and that
does not include time for the development/certification cycles. Several goals have been
set for the aviation industry to reduce its environmental impact: in 2010, the Obama
Administration set the goal of carbon neutral growth for aviation by 2020, based on
2005 emissions (FAA, 2012) and another industry commitment is to reduce aviation
emissions by 50% by 2050 (with respect to 2005 emissions) (CANSO, 2012).
Improvements in the air traffic management (ATM) system have the unique ability to
bring system wide benefits to realization. New aircraft technologies are vital to
mitigating the environmental impacts of aviation as well, but the benefits are not fully
realized for a long period of time and only impact one portion of the world fleet at a
time. Upgrades to the ATM infrastructure and the way aircraft are operated have the
potential to be far more impactful in the short-term and quicker to implement. For
example, it took 11 years to have 67% of the world adopt the Reduced Vertical
Separation Minimum (RVSM) standard (Kar, 2010).
Given the timeframe of the industry goals, quantifying the opportunity pool for ATM
improvements is important. In addition, much of the existing ATM literature focuses on
CO2 emissions only. Quantifying the ATM impact on NOx, HC, and CO is also important
due to the human health impacts these emissions have and their dependence on engine
power level (Baughcum et al., 1996), which varies during different portions of the flight.
We seek to quantify the theoretical opportunity pool for each of the above emissions
using AEIC.
38
4.2.2. Methodology
The AEIC methodology for modeling operational inefficiencies comes from Reynolds
(2008, 2009) and is covered in Section 2.5. For this study, global runs were completed
with and without inefficiency for each portion of the flight (departure, en-route, and
arrival) active.
4.2.3. Results
Table 8 contains the improvement opportunity for each species (note that CO2 and SOx
efficiencies would be equivalent to fuel burn), while Table 9 contains the contribution of
each flight phase to the overall inefficiency for each species. The total opportunity for
fuel burn reduction is approximately 6.6%. This places the fuel efficiency of the global air
traffic management system at 93.4%, which is in agreement with the Civil Air Navigation
Services Organization (CANSO) estimate of 92-94% (CANSO, 2012). It should be noted
that this case study does not account for sub-optimal cruise altitude due to ATM system,
which is estimated to be worth about 1.2% of total fuel burn (Lovegren & Hansman,
2011). The en-route phase is the largest contributor to fuel burn inefficiency.
At 6.7%, the inefficiency level for NOx is close to fuel burn, but the contribution
breakdown is different. Due to the high thrust level during the departure phase, it has a
greater contribution to NOx efficiency than the en-route phase does. Inefficiency levels
for both HC and CO (15.8% and 11.2%, respectively) are greater than the fuel burn
inefficiency level. This is due to the relatively large inefficiency in the arrival phase,
which accounts for the majority of the inefficiency for these species. One important
aspect to note is that this AEIC simulation may not fully capture the different power
levels that arrival inefficiencies occur at. This study assumes the engines are a descent
39
power level throughout the arrival portion of the flight, while actual engine thrust is
higher during maneuvers such as holding.
Table 8. ATM system improvement opportunity for different emissions
Flight Phase
Departure
Fuel Burn
-2.1%
NOx
-3.2%
HC
-0.3%
CO
-0.5%
En-Route
-3.4%
-3.1%
-1.6%
-2.1%
Arrival
-1.1%
-0.4%
-13.9%
-8.6%
Total
-6.6%
-6.7%
-15.8%
-11.2%
Table 9. Contribution of each flight phase to inefficiency for different emissions
Flight Phase
Departure
Fuel Burn
33%
NOx
48%
HC
2%
CO
5%
En-Route
51%
46%
10%
19%
Arrival
16%
6%
88%
77%
40
5. Conclusions
We have developed a methodology and open source code for calculating aircraft
emissions on a global scale in a rapid manner, with quantified uncertainty. The entire
flight, including both LTO operations and cruise, has been modeled. Sources of
uncertainty that have been accounted for include operational factors (e.g. lateral
inefficiencies and times-in-mode), scientific knowledge (e.g. emissions indices), and
model fidelity (e.g. fuel flow and drag calculations).
We estimate that the worldwide fuel burn for 2005 scheduled civil aviation operations is
approximately 180.6 Tg (90% CI: 136.1-232.9 Tg), equating to 155.5 Tg of CO2 as C (90%
CI: 117.3-200.7 Tg) and 0.108 Tg of SOx as S (90% CI: 0.080-0.142 Tg) emissions. 2.689 Tg
of NOx as NO2 (90% CI: 1.761-3.804 Tg), 0.749 Tg of CO (90% CI: 0.422-1.145 Tg), and
0.201 Tg of HC as CH4 (90% CI: 0.072-0.362 Tg) were also emitted. The largest relative
uncertainty is in HC emissions due to the uncertainty range on its emission index and its
sensitivity to engine power level.
92% of fuel burn takes place in the northern hemisphere, while 67% percent of fuel burn
occurs between 30°N and 60°N. Fuel burn within the longitude bands of 120°W - 60°W
15°W - 30°E, and 90°E to 150°E accounts for 72.6% of the global total. LTO operations
from aircraft at or near the surface (0-3000 feet AFL) make up 9.1% of global fuel burn,
while 70.6% of fuel burn occurs at cruise altitudes (>8 km).
The United States accounts for the largest portion of global fuel burn. Hong Kong and
Singapore have the high per capita fuel burn, equating to about 2.4 tonnes of CO2 per
person. 73% of fuel burn associated with the EU is due to non-LTO phases of
International flights. Countries with the greatest area (e.g. United States of America,
China, and Russia) have the highest level of absolute fuel burn within their borders,
while smaller European countries tend to have the highest fuel burn density.
41
Our global fuel burn totals are within 4% of other published inventories for the years
2004 and 2006, which is within the 90% confidence interval of our analysis. The
longitudinal, latitudinal, and altitudinal distributions of this thesis also agree well with
other inventories. To our knowledge, this inventory is the most current emissions
inventory for aviation publicly available, the only one for which the underlying code is
open source, and the only to estimate emissions uncertainty fleet-wide.
Water injection has the potential to reduce the emission of NOx during takeoff by 59.4%
with a relatively small increase in fuel burn (+1.0%). When averaged over all global
emissions, this equates to a NOx reduction of 4.4% with a fuel burn penalty of only 0.2%.
Improvements in the ATM system have the potential to reduce aviation emissions. The
total fuel burn improvement opportunity is 6.6%, while the potential improvement in
NOx emissions is 6.7%. The improvement opportunities for HC and CO are larger at
15.8% and 11.2%, respectively. This can be compared to the forecast annual growth rate
of up to 6.2% per year (ICAO, 2012)
The development of a rapid open source emissions tool allows for full scale simulations
to be used for studies where the computational time of higher fidelity models makes
them impractical, such as rapid policy analyses and uncertainty analyses. It can also be
used in scientific assessments. For example, AEIC is being applied in studies by Gilmore
et al. (2013, forthcoming) and flight-level impacts of aircraft NOx emissions on
tropospheric ozone, and by Stettler et al. (2013, forthcoming) on black carbon emissions
from aviation. The 2005 AEIC inventory has also been incorporated into the open source
atmospheric chemistry transport model GEOS-Chem.
42
[Page Intentionally Left Blank]
43
6. References
Airbus, 2012. Global Market Forecast 2012-2031. Airbus, France.
Barrett, S.R.H., Britter, R.E. & Waitz, I. A., 2010a. Global mortality attributable to aircraft
cruise emissions. Environmental science & technology. 44 (19). 7736–7742.
Barrett, S.R.H., Prather, M., Penner, J., Selkirk, H., Dopelheuer, A., Fleming, G., Gupta,
M., Halthore, R., Hileman, J., Jacobson, M., Kuhn, S., Miake-lye, R., Petzold, A., Roof, C.,
Schumann, U., Waitz, I. & Wayson, R., 2010b. Guidance on the use of AEDT Gridded
Aircraft Emissions in Atmospheric Models v2.0.
Baughcum, S.L., Begin, J. & Franco, F., 1999. Aircraft Emissions: Current Inventories and
Future Scenarios. In: D. J. G. J.E. Penner, D.H. Lister (ed.). Aviation and the Global
Atmosphere. 1999, Intergovernmental Panel on Climate Change, pp. 290–331.
Baughcum, S.L., Tritz, T., Henderson, S. & Pickett, D., 1996. Scheduled Civil Aircraft
Emission Inventories and Analysis for 1992 : Database Development and Anlaysis. NASA
Contractor Report 4700.
British Airways/IATA, 2002. Take-off at less than full power. ICAO/CAEP/Working Group
3/AEM Task Group. June 27-28, 2002.
CAA, 2009. Aircraft Engine Emissions.
CANSO, 2012. Accelerating Air Traffic Management Efficiency: A Call to Industry.
CIESIN/Columbia Univerity/CIAT, 2005. Gridded Population of the World, Version 3
(GPWv3): National Identifier Grid. Available at online:
http://sedac.ciesin.columbia.edu/data/collection/gpw-v3. [Accessed: 6 November
2012].
Daggett, D., Hendricks, R., Mahashabde, A., & Waitz, I., 2007. Water Injection—Could it
Reduce Airplane Maintenance Costs and Airport Emissions? NASA/TM—2007-213652.
Daggett, D. L., Fucke, L., Airplane, B. C., Hendricks, R. C., & Eames, D. J. H., 2010. Water
Injection on Commercial Aircraft to Reduce Airport Nitrogen Oxides. NASA/TM—2010213179.
Eurocontrol Experimental Center, 2011. User Manual for The Base Of Aircraft Data
(Bada) Revision 3.9. EEC Technical/Scientific Report No. 11/03/08-08.
44
European Union, 2012. Member states of the EU. 2012. Available at online:
http://europa.eu/about-eu/countries/index_en.htm. [Accessed: 11 December 2012].
Eyers, C.J., Norman, P., Middel, J., Plohr, M., Michot, S., Atkinson, K. & Christou, R.,
2004. AERO2k Global Aviation Emissions Inventories for 2002 and 2025.
QINETIQ/04/01113.
FAA, 2012. United States Aviation Greenhouse Gas Emissions Reduction Plan.
Gauss, M., Isaksen, I., DS, L. & Sovde, O., 2006. Impact of aircraft NOx emissions on the
atmosphere – tradeoffs to reduce the impact. Atmospheric Chemistry and Physics. 6.
1529–1548.
Hileman, J.I., Donohoo, P.E. & Stratton, R.W., 2010. Energy Content and Alternative Jet
Fuel Viability. Journal of Propulsion and Power. 26 (6). 1184–1196.
ICAO, 2007. Review of the Fleet and Operations Module (FOM) Assumptions and
Limitations.
ICAO, 2012. Forecasts of Scheduled Passenger Traffic. 2012. Available at online:
http://www.icao.int/sustainability/pages/eap_fp_forecast_longterm.aspx. [Accessed: 8
November 2012].
ICAO, 2008. ICAO Annex 16: Environmental Protection, Volume II -- Aircraft Engine
Emissions. 552.
ICAO, 2010. ICAO Environmental Report 2010. Aviation and Climate Change.
IEA/OECD, 2007. IEA Energy Balance. International Energy Agency: Paris.
Kar, R., 2010. Dynamics of implementation of mitigating measures to reduce CO2
emissions from commercial aviation. MIT.
Kim, B., Fleming, G., Lee, J., Waitz, I., Clarke, J., Balasubramanian, S., Malwitz, a, Klima,
K., Locke, M. & Holsclaw, C., 2007. System for assessing Aviation’s Global Emissions
(SAGE), Part 1: Model description and inventory results. Transportation Research Part D:
Transport and Environment. 12 (5). 325–346.
Krzyzanowski, M. & Cohen, A., 2008. Update of WHO air quality guidelines. Air Quality,
Atmosphere & Health. 1 (1). 7–13.
45
Lee, D.S., Fahey, D.W., Forster, P.M., Newton, P.J., Wit, R.C.N., Lim, L.L., Owen, B. &
Sausen, R., 2009. Aviation and global climate change in the 21st century. Atmospheric
Environment. 43 (22-23). 3520–3537.
Lee, J.J., 2005. Modeling Aviation’s Global Emissions, Uncertainty Analysis, and
Applications to Policy. MIT.
Lee, J.J., Waitz, I.A., Kim, B.Y., Fleming, G.G., Maurice, L. & Holsclaw, C.A., 2007. System
for assessing Aviation’s Global Emissions (SAGE), Part 2: Uncertainty assessment.
Transportation Research Part D: Transport and Environment. 12 (6). 381–395.
Lovegren, J. A., & Hansman, R. J., 2011. Estimation of Potential Aircraft Fuel Burn
Reduction in Cruise via Speed and Altitude Optimization Strategies. MIT.
OAG Aviation, 2005. OAG Schedules Data. Available at online:
http://www.oagaviation.com/Solutions/ AnalysisTools/Schedules/schedules.html
[Accessed: 15 July 2010].
Olsen, S.C., Wuebbles, D.J. & Owen, B., 2012. Comparison of global 3-D aviation
emissions datasets. Atmospheric Chemistry and Physics Discussions. 12 (7). 16885–
16922.
Owen, B., Lee, D.S. & Lim, L., 2010. Flying into the future: aviation emissions scenarios to
2050. Environmental science & technology. 44 (7). 2255–60.
Pham, V. Van, Tang, J., Alam, S., Lokan, C. & Abbass, H.A., 2010. Aviation emission
inventory development and analysis. Environmental Modelling & Software. 25 (12).
1738–1753.
Reynolds, T.G., 2008. Analysis of Lateral Flight Inefficiency in Global Air Traffic
Management. In: 26th Congress of International Council of the Aeronautical
Sciences/8th AIAA Aviation Technology, Integration & Operations Conference. 2008,
Anchorage, AK.
Reynolds, T.G., 2009. Development of flight inefficiency metrics for environmental
performance assessment of ATM. Traffic Management Research and Development.
Rienecker, M., Suarez, M. & Todling, R., 2008. The GEOS-5 data assimilation system:
Documentation of versions 5.0. 1, 5.1. 0, and 5.2. 0. Technical Report Series on Global
Modeling and Data Assimilation. 27.
46
Stettler, M.E.J., Eastham, S. & Barrett, S.R.H., 2011. Air quality and public health impacts
of UK airports. Part I: Emissions. Atmospheric Environment. 45 (31). 5415–5424.
Sutkus Jr., D.J., Baughcum, S.L. & DuBois, D.P., 2001. Scheduled civil aircraft emission
inventories for 1999: database development and analysis. NASA/CR-2001-211216.
United Nations, 2011. World Population Prospects: The 2010 Revision, CD-ROM Edition.
United Nations, Department of Economic and Social Affairs, Population Division.
Wilkerson, J.T., Jacobson, M.Z., Malwitz, a., Balasubramanian, S., Wayson, R., Fleming,
G., Naiman, a. D. & Lele, S.K., 2010. Analysis of emission data from global commercial
aviation: 2004 and 2006. Atmospheric Chemistry and Physics. 10 (13). 6391–6408.
Yoder, T.A., 2007. Development of aircraft fuel burn modeling techniques with
applications to global emissions modeling and assessment of the benefits of reduced
vertical separation minimums. MIT.
47